NVIDIA Investor and Competitor to AMI
AMI Labs
NVIDIA’s role makes clear that AMI is building on top of a layer controlled by a company that also wants to own the broader physical AI tooling stack. AMI can benefit from NVIDIA capital and compute access, but Cosmos already gives robotics teams prediction models, controllable simulation, and synthetic data tools, which means NVIDIA is backing the category while also selling a platform that can absorb part of AMI’s value.
-
NVIDIA publicly launched Cosmos as a physical AI platform with world foundation models for prediction, world generation, and reasoning, then paired it with Omniverse and blueprints for robot and autonomous vehicle data generation. That is close to the same developer workflow a horizontal world model company needs to monetize.
-
The overlap is not theoretical. NVIDIA named Figure AI, Agility Robotics, and Skild AI as early adopters of Cosmos. Those are exactly the kinds of robotics companies that could either buy an intelligence layer from AMI or instead build on NVIDIA’s stack plus their own deployment data.
-
This is common platform economics. NVIDIA also invested in AMI’s March 2026 seed round, but its upside does not depend on AMI becoming the only winner. More world model companies, more robot developers, and more simulation workloads all pull through demand for NVIDIA compute, software, and developer tools.
The market is heading toward a split where a few integrated robot companies own end deployments and a few infrastructure companies own the shared tooling layer beneath them. For AMI to hold durable value in that structure, it will need to become the model layer customers cannot replace with Cosmos plus their own proprietary data and application workflows.